Research ArticlePortable Multispectral System Based on Color Detector for theAnalysis of Homogeneous Surfaces

A. Martínez-Olmos ,1 P. M. Olmos ,2 M. M. Erenas ,3 and P. Escobedo 1

1Department of Electronics and Computer Technology, University of Granada, 18071 Granada, Spain2Department of Signal Theory and Communications, University Carlos III, 28911 Madrid, Spain3Department of Analytical Chemistry, University of Granada, 18071 Granada, Spain

Correspondence should be addressed to A. Martínez-Olmos; amartinez@ugr.es

Received 4 September 2018; Revised 24 October 2018; Accepted 28 November 2018; Published 10 January 2019

Academic Editor: Harith Ahmad

Copyright © 2019 A. Martínez-Olmos et al. This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work isproperly cited.

In this work, a compact affordable and portable spectral imaging system is presented. The system is intended to be employed ingeneral applications, such as material classification or determination of the concentration of chemical species together withcolorimetric sensors. The imaging device is reduced to a small digital color detector with an active area of 3 × 2mm2. Thisdevice provides a quantification of the incident emission in the form of four digital words corresponding to its averagedcomponents blue, green, red, and near infrared. In this way, the size of the image is reduced to one pixel. The wavelengthselection is carried out by means of a LED array disposed surrounding the color detector. The LEDs are selected to cover thewavelength range from 360 to 890 nm. A sequential measurement protocol is followed, and the generated data is transmitted toan external portable device via a Bluetooth link where a classification protocol is implemented in a custom-developed Android™application. The presented system has been applied in three different scenarios involving material classification, meat freshnessmonitoring, and chemical analysis. The analysis of the data using principal components shows that it is possible to find a set ofwavelengths where the classification of the samples is optimal.

1. Introduction

Imaging spectrometry was firstly defined by Goetz et al. in1985 [1] “as a major advance in remote sensing which con-sists of the acquisition of images in many narrow contiguousspectral bands throughout the visible and solar-reflectedinfrared spectral bands simultaneously”. With this techniqueit is possible to acquire a complete reflectance spectrum foreach picture element (pixel) in the image that it is consideredas a cube of information [2]. The term hyperspectral refers tothe multidimensional character of the spectral data set. Thetechnique is known as multispectral imaging when theimages are acquired at several discrete wavelengths in theconsidered range of the spectrum, usually 6 to 10 or even less[3, 4]. Imaging spectrometry is usually applied over a widerange of wavelengths covering from ultraviolet (UV) at380nm to near infrared (NIR) up to 1100nm. It combinestwo methodologies, spectroscopy and imaging. The imaging

part of this technique provides the intensity at every pixelof the image, whereas the spectrometry side generates a singlespectrum for the same element [5]. Spectral imaging wasintroduced as a technique for remote sensing. Since then, thisanalysis technique has expanded to many other fields, andnowadays it is applied in a diversity of areas such as agricul-ture, military, environment, geography, medicine, and nutri-tion, among others. [6–11].

The main components that comprise a typical spectralimaging system are lighting system, focusing optics, detec-tion system, and wavelength selector. The design of the illu-mination system is mostly determined by the method forwavelength selection. First studies on multispectral imagingsystems employed a monochrome digital camera to collectthe reflected light of the previously filtered illumination[12]. Other systems filter or disperse the reflected light toselect the desired wavelength [13, 14]. More recently, wave-length selection is directly achieved within the illumination

HindawiJournal of SensorsVolume 2019, Article ID 1519753, 8 pageshttps://doi.org/10.1155/2019/1519753

source by illuminating the sample with quasimonochromaticlight. Programmable light sources are commonly used inthese kinds of systems [15]. Nowadays, the expansion of veryeffective light-emitting diodes (LEDs) that are invariant andremarkably stable as compared with white lights has enabledthe development of LED-based multispectral imaging sys-tems [16, 17]. Conventional spectral imaging systemsinclude complicated illumination sources, delicate optics,and high-resolution cameras [18–21], so they tend to beexpensive and fragile laboratory equipment. Moreover,the system control and the data collection and processingare usually implemented in a personal computer (PC).

In a previous work, the authors presented a compactmultispectral system based on a Raspberry Pi module [22].The low cost and compact size of this camera module, facingother in the market, as well as the processing capacity of theRaspberry in a compact and portable instrument, makes thisnovel system suitable for a variety of applications. Followingthese design criteria, that is, portability, low cost, and widerange of applications, a new prototype is developed and pre-sented here. The novelty of this instrument lies in the imagedetector employed and the classification software imple-mented as an Android™ application. While the previous sys-tem, as it happens in most of the described multispectralsystems, was based on a camera for the image acquisition, asimple digital color detector is used in this work. This deviceprovides the red, green, blue, and infrared (R, G, B, and IR)components of the incident light on a small active area in aform of 16-bit digital words. This is considered as an imageof only one pixel, which can be representative of an extensiveand homogeneous area, meaning that the texture and color ofthe sample are constant on the whole surface. In this way, thecomplexity of the system and the generated data is reduced,and the color depth is improved in comparison with theimages generated with classical CMOS cameras. Theseone-pixel images are transmitted to a remote portable devicesuch as a smartphone or a tablet where the results of theclassification process are presented.

The system has been evaluated in three different scenar-ios. In the first one, a material classification has been carriedout to analyze several kinds of white sheets of paper. In thesecond experience, a package of fresh pork meat has beenmonitored during eight days. In the third experiment, the

instrument has been applied to measure the concentrationof potassium in water solutions at pH9. With a simpledimensionality reduction technique (principal componentanalysis) combined with a standard low-complexity classifi-cation tool (support vector machine), it is possible to deter-mine at which wavelengths the samples can be separatedwith high precision. The classification algorithm has beenimplemented as an Android application to be used in a smartdevice that communicates with the developed instrument viaa Bluetooth protocol.

2. System Description

As it has been exposed in the previous section, the presentedsystem has been developed as a low cost portable multispec-tral imaging system where the imaging device is a digitalcolor detector that generates 1-pixel images. These data aretransmitted through a Bluetooth protocol to a portable device(smartphone or tablet) or through a USB connection to acomputer that completes the system. In both devices, the dataare processed and presented to the user. The scheme of thedeveloped system, which is exposed in detail in the followingsections, is presented in Figure 1.

2.1. Instrument Design. The portable instrument is shown inthe picture of Figure 2(a). The printed circuit board hasdimensions of 8 × 12 cm2, and it is enclosed in a black boxof dimensions 9 × 15 × 3 cm3 to avoid external illuminationinterference. This box has an aperture facing the sensing areaof the instrument composed of the imaging device and thelight source. The instrument is placed directly on the surfaceof the sample to be analyzed; in this way, the distance to theLED array and the imaging device is always the same asdepicted in Figure 2(b).

The imaging device is the color detector modelS11059-02DT (Hamamatsu Photonics K.K., Japan), whichis an I2C interface-compatible digital detector sensitive tored (575 to 660nm, λpeak = 615 nm), green (455 to 630 nm,λpeak = 530 nm), blue (400 to 540 nm, λpeak = 460 nm), andnear infrared (700 to 885nm, λpeak = 855 nm) radiations.The incident light is directly codified into words of 16 bitsof resolution. The sensitivity and integration time can be

Sensingmodule �휇controller BT module

Externalbattery

Portable instrument

Figure 1: Scheme of the presented system.

2 Journal of Sensors

adjusted so that light measurements can be performed over awide range. This device presents a photosensitive area of0 56 × 1 22mm2.

The light source consists of a LED matrix surroundingthe color detector. The selected models are the following:VLMU1610-365-135, VLMU3100 (Vishay IntertechnologyInc., USA), LD MSVG-JGLH-46-1 (OSRAM Opto Semicon-ductors GmbH, Germany), AA3021ZGSK, APTR3216PGW,APT2012NW,APT2012SRCPRV (Kingbright Electronic Co.,Ltd., China), SML-LX15HC-RP-TR (Lumex Inc., USA),VSMG2700, and VSMF3710 (Vishay Intertechnology Inc.).The emission peaks of these LEDs are, respectively, 367, 405,455, 515, 555, 610, 655, 700, 830, and 890 nm as it can beobserved in Figure 3(a), where the normalized emissionspectra of the LEDs are depicted.

There are two LEDs of each model in the array disposedin symmetrical positions as presented in Figure 3(b). In thisway, the color detector is always placed in the middle of every

two LEDs of the same model. The distribution of the array incouples of LEDs is aimed to generate a uniform irradiancedistribution on the sample. It is known that for a two-LEDarray the irradiance is homogenous at a distance z if the LEDsare separated a maximum distance d given by [23, 24]

d = 4m + 3 ⋅ z, 1

wherem is a number that depends on the relative position ofthe LED-emitting region from the curvature center of thespherical encapsulant. The value of m is given by the angleθ1/2 defined as the view angle when irradiance is half of thevalue at 0° (value typically provided by the manufacturer):

m = −ln 2ln cos θ1/2

2

00.10.20.30.40.50.60.70.80.9

1

300 400 500 600 700 800 900 1000

Nor

mal

ized

inte

nsity

Wavelength (nm)

367 nm405 nm455 nm515 nm

555 nm610 nm655 nm

700 nm830 nm890 nm

(a)

610

nm

700

nm

367 nm

405 nm

455 nm

515 nm

555 nm

655

nm

830

nm

890

nm

(b)

Figure 3: Normalized emission spectra of the LEDs (a) and scheme of the distribution of the LEDs (b).

Color detector

LEDs

Microcontroller Bluetooth module

USB connector

(a)

Sample

Sensing module

Z = 3 cm

(b)

Figure 2: Portable instrument (a) and measurement disposition (b).

3Journal of Sensors

In the presented work, the distance to the sample is deter-mined by the height of the box (see Figure 2(b)) which is3 cm. For the selected LED, the angle θ1/2 ranges from 60 to70°, therefore the value of m in equation (2) varies from0.65 to 1. This range leads to a maximum distance betweenLEDs of 3 to 3.14 cm for a distance to the sample z = 3 cm.In the design of the LED matrix of Figure 3(b), the distanceof the LEDs in couple ranges from 2.2 cm in the middle ofthe square to 3 cm from corner to corner. These values guar-antee that the irradiance on the sample is homogeneous. Thewavelength selection is carried out activating the LEDssequentially in a growing wavelength.

All the LEDs have been biased at their typical forwardvoltage by applying a voltage of 5V to the serial combinationof every LED and a resistor of 220Ω. In this situation, theLEDs showed a stable radiance with intensity dispersionbelow 0.1%. This configuration produces different intensitiesfor different LED models. In order to generate a uniformresponse of the system, the integration time for each wave-length is defined so that the product Ie × T i (intensity ofemission multiplied by the integration time) is constant. Inthis way, the integration time associated with a wavelengthwith a low-intensity emission is higher than the time corre-sponding to a wavelength generated by a high-intensityLED. These integration times range from 35 to 700ms.

The microcontroller used in this design is the modelPIC18F2550 (Microchip Technology Inc., USA). This deviceintegrates a full speed USB 2.0 (12Mbit/s) interface that hasbeen used for calibration purposes. The microcontroller acti-vates the LEDs of the light source and receives the output ofthe digital color detector. It transmits these results to anexternal computer through the USB port or to a remotedevice via a Bluetooth connection. To implement this wire-less communication, a Bluetooth module RN-42 (MicrochipTechnology Inc.) is included in the design. This module com-municates with the microcontroller through a two-wire serialprotocol and integrates a small antenna.

2.2. Android Application. A user-friendly Android applica-tion was developed to use a smartphone as the externalreader and processing unit of the multispectral imagingdevice. The instrument is connected to the smartphone forbidirectional data transmission using the Bluetooth interface.The chosen integrated development environment (IDE) tocode the application was Android Studio 3.1.3. The applica-tion was designed and tested against API 24 (Android 7.0)using a Samsung smartphone model Galaxy S7, although itsupports different Android versions as the lowest API levelcompatible with the application is API 18 (Android 4.3).

The application user interface consists of two screens.The first one, shown in Figure 4(a), allows a generic use ofthe multispectral imaging device. The user can choosebetween a sequential measurement throughout all the fre-quencies or the selection of a particular wavelength by meansof a slider control. After the measurement is done, the resultsof the R, G, B, and IR components for each wavelengthcan be consulted in a plain text report that is saved inthe internal memory of the smartphone. Figure 4(b) showsthe second user interface of the application, where the user

can choose among the three different classification scenariosin which the system has been evaluated. The processingalgorithm is changed accordingly to the chosen scenario,and the final results of the measurement are directly dis-played on the screen.

3. Results and Discussion

The presented system has been applied in three different sce-narios with very diverse objectives. The aim is to prove theapplicability of this instrument in a wide range of fields.

Automatic classification is one of the main applicationsof spectral imaging [25, 26]. In the first experience, the sys-tem has been used to analyze several kinds of white paperwith the objective of developing a classification algorithmfrom the generated data (Scenario I).

In the second experience, a package of fresh pork meathas been monitored during 8 days while it was stored at4°C (Scenario II). It is known that the packaged meat isaffected by the microbial activity, which is the main respon-sible for the food spoilage [27]. Although the external appear-ance of the meat might not be altered to the naked eye, thisbacterial growth, which increases with the storage time, pro-duces quality degradation [28]. The objective of this experi-ence is to analyze the packaged meat daily in order todevelop a prediction algorithm that is able to estimate thestorage time of packaged meat and, therefore, the status ofthe content.

In the third experiment, the instrument has been appliedto measure the concentration of potassium in water solutionsat pH9 (Scenario III). For this purpose, a potassium sensitivemembrane has been used. The reagents used to prepare thesensing membrane were 0.8mg of dibenzo18-crown-6-ether

(a) (b)

Figure 4: User interfaces of the developed Android multispectralimaging application for (a) a generic use of the instrument and (b)a specific scenario-based use.

4 Journal of Sensors

(DB18C6) as ionophore, 1.3mg of 1,2-benzo-7-(diethyla--mino)-3-(octadecanoylimino) phenoxazine (liphophilizedNile Blue) as lipophilic pH indicator, 63.0mg ofo-nitrophenyloctylether (NPOE) as plasticizer, 1.1mg ofpotassium tetrakis (4-chlorophenyl)-borate (TCPB) aslipophilic salt and 26.0mg of polyvinyl chloride (PVC) aspolymer, all dissolved in 1mL of tetrahydrofurane (THF)[29, 30]. Once dissolved, 60μL of the sensing cocktail is

dropped by spin coating on a Mylar sheet, obtaining inthis way a round-shape sensing membrane whose colorchanges depending on the potassium concentration whenit is introduced for 3 minutes in the sample, working inthe range from 10-5 to 0.95M of potassium.

In each of the three scenarios, each measurement is afour-dimensional vector xλ, where λ is the LED-emittingwavelength. We consider ten different values of the emitting

2D PCA projection (700 nm)

Prin

cipa

l com

pone

nt 2

White paperWhite paperboardBrown paperboardGlossy paper

Card paperboardWhite napkinRecycled paper

Principal component 1

(a)

2D PCA projection (405 nm)

Prin

cipa

l com

pone

nt 2

0 days1 days2 days

3 days4 days5 days

6 days7 days8 days

Principal component 1

(b)

2D PCA projection (515 nm)

Prin

cipa

l com

pone

nt 2

−6.0−5.0−4.5−4.0

−3.5−3.0−2.5

−2.0−1.00.0

Principal component 1

(c)

Figure 5: Two-dimensional projection with PCA of the dataset in Scenarios I (a), II (b), and III (c) at three wavelengths. Legend in (c) refers tologarithmic pH values.

5Journal of Sensors

wavelength (see Figure 2) and a total of six measurements perwavelength. Figures 5(a)–5(c) shows the resulting data pro-jected in a two-dimensional space using principal componentanalysis (PCA) for representative wavelengths where the datais visually separated among classes. Data has been standard-ized to have zero mean. In Figure 5(a), the results forScenario I when λ = 700 nm are shown. In Figure 5(b),Scenario II when λ = 405 nm is depicted. Figure 5(c) showsScenario III when λ = 515 nm. In all cases, data points corre-sponding to the same class (paper, days of storage, or potas-sio concentration) cluster together, reasonably separatedaway (compared to the cluster variance) from data corre-sponding to other classes.

To assess the performance of a classifier that uses thelow-dimensional measure embedding described in Figure 6,we compare the average classification test error rate for eachwavelength for a linear support vector machine (SVM) [31]when the 30% of the data is left out for test. Each experimentis repeated 1000 times for different train/test partitions. Theresults are shown by means of a box plot, where the dashedcentral line shows the median of the empirical distributionof the error along the 1000 repetitions of the experiment,the box boundaries represent the first and 3 quartiles of theerror, and the lines outside the boxes represent the extremeerror values in the sample. It can be observed that the errorrate heavily depends on the wavelength, but for every casewe can find a wavelength λ for which the error mode is closeto zero. Further, only results for Scenarios II and III arepresented, as for Scenario I the error rate for all wavelengthsis zero (i.e., a linear classifier perfectly separates the availabledata no matter the choice of the training set). For Scenario II,the average error rate is minimized at λ = 515 nm (2.25%),and for Scenario III the average error rate is minimizedat λ = 655 nm (0.4%).

Further, the normalized inverse distance to the SVMdecision boundary can be used as an approximation to

the classifier’s probabilistic confidence [32]. In the experi-ments summarized in Figure 6, it is numerically observedthat the confidence in the right class is typically 10 timeslarger than for the rest of the classes. For instance, in Sce-nario II, the average confidence in the right class is 0.567,while for the rest of the classes the average confidence isonly 0.042.

The results described above demonstrate that the pro-posed portable multispectral imaging system is able toprovide discriminative measurements using only four lightbands, and hence it can be used for cheap low-complexitydetection device for a wide set of potential industrialapplications.

The previous experiments have been repeated using theCMOS camera of a smartphone instead of the color detector,maintaining the same geometry presented in Figure 2. Sincethe analyzed samples are homogeneous in texture and color,only 1 pixel of the photography provided by the camera isconsidered. In this scenario, the difference in the two sets ofexperiments, the one carried out with the color detectorand the other with the CMOS camera, is only the color depth(16 bits for the color detector and 8 bits for the CMOS cam-era). The prediction errors generated with both systems arepresented in Table 1. As it can be observed, the error ratesin the prediction where the analysis is carried out with theCMOS camera are much higher than the ones obtained withthe original system.

4. Conclusions

In this work, we have presented a prototype of a multispectralimaging system for general purposes. It is based on the use ofa high-resolution digital color detector as the imaging deviceand a LED panel for the wavelength selection. The colordetector generates one-pixel images in a four-dimensionspace (R, G, B, and IR) that are considered representative of

Erro

r rat

e

0.8

0.7

0.6

0.5

0.4

0.3

0.2

0.1

0.0

367 405 455 515 555 610 655 700 830 890 Wavelength (nm)

(a)

367 405 455 515 555 610 655 700 830 890 Wavelength (nm)

Erro

r rat

e

0.6

0.5

0.4

0.3

0.2

0.1

0.0

(b)

Figure 6: Classification performance for Scenario II (a) and Scenario III (b) using a linear SVM classifier trained over 70% of theavailable data.

6 Journal of Sensors

an extensive homogeneous area. These images are transmit-ted to a remote smart device through a Bluetooth wirelessconnection. A custom application for Android operative sys-tem has been developed for the acquiring and processing ofthe images. This scheme implies a simplification of the tradi-tional multispectral imaging system based on a complexcamera not only in the hardware requirements but also interms of signal processing. The result has been a compactand portable field measuring instrument that is easy to usefor any nontrained user.

The feasibility of the prototype has been proved in threedifferent scenarios. In the first one, the analysis of differentwhite paper samples has been carried out. In the secondone, the external appearance of packaged pork meat has beenmonitored during 8 consecutive days. In the last experience,the system has been applied for the determination of potas-sium concentration by the analysis of a potassium-sensitivemembrane. The generated data in each experiment have beenstudied by multivariate analysis techniques such as principalcomponent, finding that in every case it is possible to select awavelength for which the samples can be separated with highprecision. Therefore, we have proved that the presented sys-tem offers broad application possibilities including automaticsamples classification, monitoring of the status of storedfood, or colorimetric determination of analytes in solution.

Data Availability

The data used to support the findings of this study areavailable from the corresponding author upon request.

Conflicts of Interest

The authors declare that they have no conflicts of interest.

Acknowledgments

This work was supported by project CTQ2016-78754-C2-1-Rfrom the Spanish Ministry of Economy and Competitiveness.P. Escobedo wants to thank the Spanish Ministry ofEducation, Culture and Sport (MECD) for a predoctoral

grant (FPU13/05032). This study was partially supportedby the Unidad de Excelencia de Química aplicada a bio-medicina y medioambiente, Universidad de Granada.

References

[1] A. F. H. Goetz, G. Vane, J. E. Solomon, and B. N. Rock, “Imag-ing spectrometry for Earth remote sensing,” Science, vol. 228,no. 4704, pp. 1147–−1153, 1985.

[2] Y. Garini, I. T. Young, and G. McNamara, “Spectral imaging:principles and applications,” Cytometry Part A, vol. 69A,no. 8, pp. 735–747, 2006.

[3] F. Kazemzadeh, S. A. Haider, C. Scharfenberger, A. Wong, andD. A. Clausi, “Multispectral stereoscopic imaging device:simultaneous multiview imaging from the visible to thenear-infrared,” IEEE Transactions on Instrumentation andMeasurement, vol. 63, no. 7, pp. 1871–1873, 2014.

[4] W. Huang, J. Li, Q. Wang, and L. Chen, “Development of amultispectral imaging system for online detection of bruiseson apples,” Journal of Food Engineering, vol. 146, pp. 62–71,2015.

[5] N. Gat, “Imaging spectroscopy using tunable filters: a review,”in Proceedings Volume 4056, Wavelet Applications VII, pp. 50–64, Orlando, FL, USA, April 2000.

[6] Y. Zhao, C. Yi, S. G. Kong, Q. Pan, and Y. Cheng, Multi-BandPolarization Imaging and Applications, Springer-Verlag,Berlin Heidelberg, 2016.

[7] Y. Ma, R. Li, G. Yang, L. Sun, and J. Wang, “A research on thecombination strategies of multiple features for hyperspectralremote sensing image classification,” Journal of Sensors,vol. 2018, Article ID 7341973, 14 pages, 2018.

[8] Q. Huang, Q. Chen, H. Li, G. Huang, Q. Ouyang, and J. Zhao,“Non-destructively sensing pork’s freshness indicator usingnear infrared multispectral imaging technique,” Journal ofFood Engineering, vol. 154, no. 116, pp. 69–75, 2015.

[9] M. S. Braga, R. F. V. V. Jaimes, W. Borysow, O. F. Gomes, andW. J. Salcedo, “Portable multispectral colorimeter for metallicion detection and classification,” Sensors, vol. 17, no. 8, article1730, 2017.

[10] H. Steiner, S. Sporrer, A. Kolb, and N. Jung, “Design of anactive multispectral SWIR camera system for skin detection

Table 1: Comparison of prediction error rate in the 3 different scenarios when the analysis is carried out using the color detector and theCMOS camera.

Lambda (nm)Error rate Scenario I Error rate Scenario II Error rate Scenario III

Color detector CMOS camera Color detector CMOS camera Color detector CMOS camera

367 0.000 0.000 0.121 0.882 0.051 0.757

405 0.000 0.000 0.034 0.681 0.027 0.379

455 0.000 0.000 0.042 0.534 0.040 0.439

515 0.000 0.000 0.022 0.441 0.006 0.146

555 0.000 0.162 0.030 0.564 0.011 0.311

610 0.007 0.210 0.530 0.822 0.020 0.226

655 0.000 0.000 0.434 0.544 0.006 0.268

700 0.000 0.153 0.204 0.628 0.023 0.380

830 0.000 0.000 0.252 0.407 0.259 0.723

890 0.000 0.000 0.290 0.337 0.423 0.622

7Journal of Sensors

and face verification,” Journal of Sensors, vol. 2016, Article ID9682453, 16 pages, 2016.

[11] H. Liu, S. H. Lee, and J. S. Chahl, “A multispectral 3-D visionsystem for invertebrate detection on crops,” IEEE SensorsJournal, vol. 17, no. 22, pp. 7502–7515, 2017.

[12] D. Saunders and J. Cupitt, “Image processing at the nationalgallery: the VASARI project,” National Gallery TechnicalBulletin, vol. 14, pp. 72–85, 1993.

[13] H. Liang, “Advances in multispectral and hyperspectral imag-ing for archaeology and art conservation,” Applied Physics A:Materials Science & Processing, vol. 106, no. 2, pp. 309–323,2012.

[14] H. Erives and N. B. Targhetta, “Implementation of a 3-Dhyperspectral instrument for skin imaging applications,” IEEETransactions on Instrumentation and Measurement, vol. 58,no. 3, pp. 631–638, 2009.

[15] S. Tominaga, “CIC@20: multispectral imaging,” in Color andImaging Conference, 20th Color and Imaging Conference FinalProgram and Proceedings, Society for Imaging Science andTechnology, Los Angeles, CA, USA, January 2012.

[16] Y. Gong, D. Zhang, P. Shi, and J. Yan, “High-speed multispec-tral iris capture system design,” IEEE Transactions on Instru-mentation and Measurement, vol. 61, no. 7, pp. 1966–1978,2012.

[17] R. Shrestha and J. Y. Hardeberg, “Multispectral imaging usingLED illumination and an RGB camera,” in Color and ImagingConference, 21st Color and Imaging Conference Final Programand Proceedings, Society for Imaging Science and Technology,January 2013.

[18] J. Park, M. H. Lee, M. D. Grossberg, and S. K. Nayar, “Multi-spectral imaging using multiplexed illumination,” in 2007IEEE 11th International Conference on Computer Vision, Riode Janeiro, Brazil, October 2007.

[19] S. Tominaga and T. Horiuchi, “Spectral imaging by synchro-nizing capture and illumination,” Journal of the Optical Societyof America A, vol. 29, no. 9, pp. 1764–1775, 2012.

[20] T. H. Kim, H. J. Kong, T. H. Kim, and J. S. Shin, “Design andfabrication of a 900–1700 nm hyper-spectral imagingspectrometer,” Optics Communication, vol. 283, no. 3,pp. 355–361, 2010.

[21] M. Zucco, V. Caricato, A. Egidi, and M. Pisani, “A hyperspec-tral camera in the UVA band,” IEEE Transactions on Instru-mentation and Measurement, vol. 64, no. 6, pp. 1425–1430,2015.

[22] N. López-Ruiz, F. Granados-Ortega, M. A. Carvajal, andA. Martínez-Olmos, “Portable multispectral imaging systembased on Raspberry Pi,” Sensor Review, vol. 37, no. 3,pp. 322–329, 2017.

[23] I. Moreno, M. Avendaño-Alejo, and R. I. Tzonchev, “Design-ing light-emitting diode arrays for uniform near-fieldirradiance,” Applied Optics, vol. 45, no. 10, pp. 2265–2272,2006.

[24] A. J.-W. Whang, Y.-Y. Chen, and Y.-T. Teng, “Designinguniform illumination systems by surface-tailored lens andconfigurations of LED arrays,” Journal of Display Technology,vol. 5, no. 3, pp. 94–103, 2009.

[25] P. Jonsson, J. Casselgren, and B. Thörnberg, “Road surfacestatus classification using spectral analysis of NIR cameraimages,” IEEE Sensors Journal, vol. 15, no. 3, pp. 1641–1656,2015.

[26] C. Nansen, M. Kolomiets, and X. Gao, “Considerations regard-ing the use of hyperspectral imaging data in classifications offood products, exemplified by analysis of maize kernels,” Jour-nal of Agricultural and Food Chemistry, vol. 56, no. 9,pp. 2933–2938, 2008.

[27] Z. Fang, Y. Zhao, R. D. Warner, and S. K. Johnson, “Active andintelligent packaging in meat industry,” Trends in Food Scienceand Technology, vol. 61, pp. 60–71, 2017.

[28] I. M. Pérez de Vargas-Sansalvador, M. M. Erenas,D. Diamond, B. Quilty, and L. F. Capitan-Vallvey, “Waterbased-ionic liquid carbon dioxide sensor for applications inthe food industry,” Sensors and Actuators B: Chemical,vol. 253, pp. 302–309, 2017.

[29] M. M. Erenas, O. Piñeiro, M. C. Pegalajar, M. P. Cuellar, I. deOrbe-Payá, and L. F. Capitán-Vallvey, “A surface fit approachwith a disposable optical tongue for alkaline ion analysis,”Analytica Chimica Acta, vol. 694, no. 1-2, pp. 128–135, 2011.

[30] M. M. Erenas, K. Cantrell, J. Ballesta-Claver, I. De Orbe-Payá,and L. F. Capitán-Vallvey, “Use of digital reflection devices formeasurement using hue-based optical sensors,” Sensors andActuators B: Chemical, vol. 174, pp. 10–17, 2012.

[31] C. M. Bishop, Pattern recognition and machine learning(Information Science and Statistics), Springer-Verlag Berlin,Heidelberg, 2006.

[32] Y. Grandvalet, J. Mariethoz, and S. Bengio, “A probabilisticinterpretation of SVMs with an application to unbalancedclassification,” in Advances in Neural Information ProcessingSystems 18 (NIPS), MIT press, 2005.

8 Journal of Sensors

International Journal of

AerospaceEngineeringHindawiwww.hindawi.com Volume 2018

RoboticsJournal of

Hindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com Volume 2018

Active and Passive Electronic Components

VLSI Design

Hindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com Volume 2018

Shock and Vibration

Hindawiwww.hindawi.com Volume 2018

Civil EngineeringAdvances in

Acoustics and VibrationAdvances in

Hindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com Volume 2018

Electrical and Computer Engineering

Journal of

Advances inOptoElectronics

Hindawiwww.hindawi.com

Volume 2018

Hindawi Publishing Corporation http://www.hindawi.com Volume 2013Hindawiwww.hindawi.com

The Scientific World Journal

Volume 2018

Control Scienceand Engineering

Journal of

Hindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com

Journal ofEngineeringVolume 2018

SensorsJournal of

Hindawiwww.hindawi.com Volume 2018

International Journal of

RotatingMachinery

Hindawiwww.hindawi.com Volume 2018

Modelling &Simulationin EngineeringHindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com Volume 2018

Chemical EngineeringInternational Journal of Antennas and

Propagation

International Journal of

Hindawiwww.hindawi.com Volume 2018

Hindawiwww.hindawi.com Volume 2018

Navigation and Observation

International Journal of

Hindawi

www.hindawi.com Volume 2018

Advances in

Multimedia

Submit your manuscripts atwww.hindawi.com